AI Interaction Tool - MCP Server

Modern AI interaction tool with advanced UI and powerful features for Model Context Protocol (MCP)

🚀 Core Features

🎯 Main Capabilities

• Interactive UI Popup for content input and conversation control

• File/Folder Attachment from workspace with validation and preview

• 🖼️ Image Attachment System with drag & drop, multi-image support

• Multi-language Support (English/Vietnamese)

• Maximum Cognitive Power activation for peak AI performance

• Tag-based Output Format integrated with system prompt rules

• Workspace-aware Path Processing for cross-project compatibility

🔧 New in v2.2.0 (Latest)

• 🖼️ Image Attachment Support with drag & drop functionality

• 🛡️ Security Enhanced - secure path storage in user_images directory

• 💾 Persistent Image State - checkbox state saves correctly

• 🎯 Multi-image Management - attach, preview, and remove multiple images

• 🔄 Database Auto-cleanup - automatic image cleanup when disabled

🔧 Previous v2.1.0

• Enhanced UI/UX with modern PyQt5 interface

• Structured Tag-based Output for perfect AI agent integration

• Debounce Configuration with smart auto-save mechanisms

• Cursor IDE Integration with comprehensive setup guide

📋 Installation & Setup Guide

📥 Step 1: Clone Repository

🐍 Step 2: Install Python

• Requirement: Python 3.8+

• Download from python.org

• Or use package manager:

  # Windows with Chocolatey choco install python # macOS with Homebrew brew install python # Ubuntu/Debian sudo apt update && sudo apt install python3 python3-pip

📦 Step 3: Install Dependencies

⚙️ Step 4: Configure MCP Server in Claude Desktop

Add the following configuration to Claude Desktop config file:

Config file paths:

• Windows: %APPDATA%\Claude\claude_desktop_config.json

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

• Linux: ~/.config/claude/claude_desktop_config.json

Configuration content:

⚠️ Important: Replace E:/MCP-servers-github/AI-interaction/mcp_server.py with the absolute path to mcp_server.py on your system.

🧠 Step 5: Configure AI Agent Rules (REQUIRED)

For proper AI agent operation with ai_interaction tool, you MUST setup custom instructions:

📋 How to Add Custom Instructions:

1. Open Claude Desktop or access Claude web interface

2. Find "Custom Instructions" or "Add custom instructions" in settings

3. Copy entire content from one of the rule files:

   • 🇻🇳 Vietnamese: rule_for_ai_VI.txt

   • 🇺🇸 English: rule_for_ai_EN.txt

4. Paste into custom instructions field and save

🎯 Why This is Necessary:

• ✅ Behavioral Framework: Rules define how AI agent processes ai_interaction output

• ✅ Thinking Protocols: Activates high-level thinking patterns for quality responses

• ✅ Ultra-Enhancement Modes: 10 cognitive modes for maximum performance

• ✅ Tag Processing: Reads and processes control tags like <AI_INTERACTION_CONTINUE_CHAT>

• ✅ Continue Logic: Auto-recall ai_interaction when continue_chat=true

📁 Rule Files Location:

⚡ Quick Setup Commands:

🚀 Step 6: Configure Cursor IDE (Recommended)

Cursor is the recommended IDE for AI development with this tool:

📋 Cursor Setup Steps:

1. Download Cursor: https://cursor.sh/

2. Install and open workspace: Open AI-interaction folder

3. Configure MCP in Cursor:

   • Open Command Palette (Cmd/Ctrl + Shift + P)

   • Search "Configure MCP Servers"

   • Add AI_interaction server config

4. Setup custom instructions:

   • Copy content from rule_for_ai_VI.txt or rule_for_ai_EN.txt

   • Paste into "Custom Instructions" field in custom mode Agent:

🎯 Cursor Advantages:

• ✅ Native MCP Support: Built-in integration with MCP servers

• ✅ AI-First IDE: Optimized for AI development workflows

• ✅ Real-time Suggestions: Context-aware code completion

• ✅ Advanced Debugging: Enhanced debugging for MCP tools

• ✅ Performance: Faster than traditional IDEs for AI projects

🚀 Step 7: Launch and Test

!!! -----> In your terminal: python E:\MCP-servers-github\AI-interaction\main.py --ui

⚠️ Important: Replace E:/MCP-servers-github/AI-interaction/mcp_server.py with the absolute path to mcp_server.py on your system. ---> AUTO SHOW UI:

1. Restart Claude Desktop/Cursor after configuring MCP server

2. Test connection by calling ai_interaction tool

3. Test UI popup to verify functionality

4. Validate rule integration through AI agent responses

📦 Package Structure

🎮 Usage Guide

Available Tools in MCP Server

1. ai_interaction: Main Interactive Tool

• Function: Creates UI popup for user input with file/image attachment

• Output: Structured tag-based format with image support

• Integration: Perfect integration with system prompt rules

• Use cases:

  • Input complex content with formatting

  • Attach files/folders from workspace

  • 🖼️ Attach images with drag & drop functionality

  • 📷 Multi-image support with preview and management

  • Control AI thinking modes and reasoning levels

Basic Usage Examples

🖼️ Image Attachment Features

📷 Core Image Capabilities

• Drag & Drop Support: Drag images directly into the UI

• Multi-image Management: Attach, preview, and remove multiple images

• Format Support: PNG, JPG, JPEG, GIF, BMP, WEBP

• Secure Storage: Images stored safely in user_images/ directory

• Base64 Encoding: Automatic conversion for AI processing

• Preview System: Click images to view larger versions

• Persistent State: Save images option with checkbox persistence

🎯 How to Use Image Attachment

1. Attach Button: Click "📷 Attach Images" to select files

2. Drag & Drop: Drag images from file explorer directly to UI

3. Paste Support: Paste images from clipboard (Ctrl+V)

4. Multiple Images: Attach as many images as needed

5. Remove Images: Click X button on individual image previews

6. Clear All: Use "🗑️ Clear Images" to remove all at once

7. Save Toggle: Check/uncheck "Save images" to control persistence

🛡️ Security & Privacy

• Local Only: All images stored locally in user_images/

• No External Access: No uploads or external connections

• Relative Paths: Only relative paths stored in config for security

• User Control: Users control what images to attach and save

• Auto-cleanup: Images automatically cleaned when save disabled

Output Format

AI Interaction Tool uses clean tag-based format:

Note: When images are attached, they are automatically converted to base64 format and included in the response for AI processing.

🔧 Troubleshooting

Common Issues

1. "Command not found" error

   • Check Python is installed and in PATH

   • Verify absolute path in MCP config

2. "Module not found" error

   • Run pip install -r requirements.txt

   • Check virtual environment if using one

3. UI not displaying

   • Ensure PyQt5 is installed correctly

   • Check display settings and desktop environment

4. File attachment not working

   • Verify file permissions and access rights

   • Check workspace path configuration

5. 🖼️ Image attachment issues

   • Ensure PyQt5 is properly installed for image processing

   • Check user_images/ directory permissions

   • Verify image formats: PNG, JPG, JPEG, GIF, BMP, WEBP supported

   • Clear config if images not loading: Remove last_attached_images from config.json

6. MCP Connection Issues in Cursor

   • Verify MCP server configuration in Cursor settings

   • Check process running with ps aux | grep mcp_server

   • Restart Cursor after config changes

Debug Mode

To debug issues, run server directly:

For Cursor debugging:

🔄 Version History

• v2.2.0 (Latest): 🖼️ Image Attachment System - Complete image support with drag & drop, multi-image management, security enhancements, and persistent state

• v2.1.0: Enhanced UI/UX, Cursor IDE integration, Debounce config system

• v2.0.0: Refactored architecture with modern PyQt5 UI

• v1.x: Core functionality and basic features

🎯 v2.2.0 Detailed Changes:

• ✅ Image Attachment UI: Full drag & drop interface with preview system

• ✅ Multi-format Support: PNG, JPG, JPEG, GIF, BMP, WEBP compatibility

• ✅ Security Hardening: Secure path storage, local-only processing

• ✅ Database Management: Auto-cleanup, persistent storage, state management

• ✅ UX Improvements: Click-to-enlarge, remove buttons, checkbox persistence

• ✅ Performance: Optimized image loading with base64 conversion

• ✅ Bug Fixes: Checkbox state persistence, config loading issues resolved

🎯 Integration Workflow & System Architecture

🔄 Complete Integration Flow:

🧠 Cognitive Enhancement System:

• Standard Mode: High-level thinking with 1+ thinking blocks

• Ultra-Enhancement Mode: 10 breakthrough cognitive modes simultaneously

  • Quantum Cognitive Mode

  • Meta-Cognitive Orchestration

  • Expert Persona Simulation

  • Time-Dilated Processing

  • Systems-Level Integration

  • Psychological Priming Mode

  • Maximum Cognitive Resource Allocation

  • Adversarial Self-Testing Mode

  • Obsessive Quality Standards

  • Breakthrough Innovation Mode

📊 Output Tag System:

💡 Advanced Features & Best Practices

🎨 UI/UX Enhancements:

• Responsive Design: Adaptive sizing with minimum 800x700 resolution

• Multi-language Support: Seamless EN/VI switching with persistent config

• Modern PyQt5 Styling: Semantic color system with button properties

• File Drag-Drop: Intuitive file attachment with validation

• Context Menu: Right-click operations for file management

• Debounce Saving: Smart config persistence with QTimer optimization

🔧 Technical Specifications:

• Python: 3.8+ required with PyQt5 dependencies

• Memory: Minimum 512MB RAM for UI components

• Storage: ~50MB for tool installation and config

• Platform: Cross-platform (Windows/macOS/Linux) with native styling

• Performance: Event-driven architecture with minimal CPU usage

📈 Performance Optimization:

• Lazy Loading: Components load only when needed

• Efficient Config: JSON-based with automatic compression

• Resource Management: Proper cleanup and memory management

• Caching Strategy: Workspace state persistence for faster startup

🛡️ Security & Privacy

🔒 Security Features:

• Local Processing: All file operations are local only, no uploads

• Path Validation: Robust security checks for file access

• Sandboxed Execution: Tool runs in controlled environment

• No Data Collection: Zero telemetry or external data transmission

🔐 Privacy Protection:

• Config Encryption: Local config with secure storage options

• File Access Control: User-controlled file attachment permissions

• Workspace Isolation: Project boundaries are enforced

• Audit Trail: Optional logging for security monitoring

🌟 System Requirements & Compatibility

💻 Minimum System Requirements:

🎯 Recommended Setup:

🔧 Compatibility Matrix:

🤝 Contributing

Note: This is a private repository. Only the owner has push access.

For suggestions or issues:

1. Create detailed issue reports

2. Provide reproduction steps

3. Include system information

4. Attach relevant logs or screenshots

📚 Documentation & Resources

📖 Documentation Files:

• rule_for_ai_VI.txt - Vietnamese agent behavior rules

• rule_for_ai_EN.txt - English agent behavior rules

• SYSTEM_PROMPT_Claude-4-sonnet-max.txt - Full system prompt example

• pyproject.toml - Project configuration and dependencies

🔗 Useful Links:

• Model Context Protocol (MCP) Documentation

• Claude Desktop Official Guide

• Cursor IDE Official Site

• PyQt5 Documentation

• Python Virtual Environments

💡 Related Projects:

- https://github.com/KhaiHuynhVN/mcp-server-agent-comm

📄 License & Legal

📜 License:

⚖️ Legal Notes:

• Tool complies with local processing requirements

• No personal data collection

• Respects user privacy and data sovereignty

• Compatible with enterprise security policies

🎯 Special Thanks:

• Model Context Protocol team for standardized interface

• Claude Desktop integration ecosystem

• Cursor IDE team for AI-first development tools

• Open source Python community

• Beta testers and early adopters

🔥 Inspiration:

Project inspired by the need for seamless AI interaction tools with modern UX principles and professional-grade architecture.

🚀 Happy Coding with AI Interaction Tool!

For support, issues, or feature requests, please open an issue on the GitHub repository.